386 



THE GRAMMAR OF SCIENCE 



polygons for (iii.) the whole series of capsules, (i.) and (ii.) 

 the special individual poppies given on p. 383. We note at 

 once how the individual poppy has a different type from the 

 race, and how it is less variable than the race. Further, 

 we notice how it would be possible, if individual types 

 are in whole or part inherited, by " selecting " these two 

 individuals to create two new races of poppy differing 

 from each other and from the original race in type. The 

 manner in which the laws of inheritance enable us to do 

 this will be illustrated in the next chapter. 



But this graphical exhibition of the distribution of 

 deviations, while useful for many purposes, does not 

 provide the numerical value of the variation we are in 

 search of 



Now suppose we took a light bar, graduated uniformly 

 like that represented in the diagram below, and placed it 



(P 



@ 



w i^m 



^ 



® 



Fig. 27. 



upon a rough pivot at the point of the scale representing 

 the mean of the system of stigmatic bands ; let us further 

 sling from this bar at the corresponding points of the 

 scale, weights proportional to the frequencies of each set 

 of bands 5 to 16. Then if this bar be set rotating on the 

 given rough pivot at a given speed, friction will bring it 

 to rest in a certain time. Now the greater the concentra- 

 tion of weights about the pivot, the sooner the bar comes 

 to rest ; the farther out from the pivot the weights are, the 

 longer it takes to come to rest. In other words, the 

 time the bar takes to come to rest is a measure such as 

 we are seeking of the concentration or scattering of the 

 weights along the range. 



Now physicists tell us that this time is proportional to the 

 square of a certain quantity termed the spin- or swing- radius, 

 and which I will denote by the Greek letter a. a" is then 

 shown to be the mean of the squares of all the individual 



